Skip to main content
skill CX Codex

presentation-master — Claude Code skill for presentations

presentation-master guides presentation planning, slide visuals, and quality checks. Use it to shape a talk, training deck, or executive update before building the slides.

MIT 3 installs ★ 9 updated 2026-03-14

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add krishagel/geoffrey --skill presentation-master -a claude-code
Open on GitHub

What presentation-master does

Start with a topic, audience, and duration. The skill proposes story structures, then drafts a slide outline and a visual plan for approval. Before image generation, it presents the proposed visuals and estimated cost. You approve each checkpoint before it moves on.

PPTX is the current output format. The skill builds slides with the existing pptx skill, PptxGenJS, or html2pptx.js; image creation uses image-gen. Its checks cover readability and design rules, including a critical failure for fonts below 30pt or contrast under 4.5:1. The pptx skill is needed for PPTX creation.

The result includes the presentation and a quality report. Google Slides and Canva are described as later stages, so the current workflow is centered on PPTX.

What it helps with

  • Plan a deck around its audience and duration
  • Turn a topic into a slide-by-slide story
  • Choose visual treatments for individual slides
  • Check a deck for readability and design rule violations
  • Review a presentation quality score and report

Example prompts for the agent

  • Create a 20-minute presentation about API security for a technical audience.
  • Draft an executive product deck outline and pause for my approval.
  • Check this PPTX for font size, contrast, and text-heavy slides.

How to install presentation-master

1Skills CLI (Vercel)

npx skills add krishagel/geoffrey --skill presentation-master -a claude-code

2Manually: put the files in place

~/.claude/skills/presentation-master/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/presentation-master/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about presentation-master

How do I install presentation-master in Claude Code?

Run npx skills add krishagel/geoffrey --skill presentation-master -a claude-code. The skill file is at ~/.claude/skills/presentation-master/SKILL.md.

What does presentation-master do?

It guides presentation planning, gets approval for the outline and visual direction, builds a PPTX, and checks it against design rules.

What does the presentation workflow require?

PPTX creation uses the existing pptx skill, while image generation uses image-gen. PPTX is the current output format.

Source text

SKILL.md · MIT
---
name: presentation-master
description: World-class presentation creation embodying principles from Garr Reynolds, Nancy Duarte, Guy Kawasaki, Seth Godin, and TED
triggers:
  - "create presentation"
  - "make presentation"
  - "build presentation"
  - "design deck"
  - "make slides"
  - "create slides"
  - "build deck"
  - "presentation for"
  - "slides for"
allowed-tools: Read, Write, Bash, WebSearch, Skill, AskUserQuestion
version: 1.0.0
---

# Presentation Master

World-class presentation creation skill that embodies best practices from presentation masters and adapts to your needs.

## Philosophy

**Key Innovation**: This isn't just a presentation generator—it's a presentation *coach* that teaches world-class design while creating slides.

Principles from:
- **Garr Reynolds** (Presentation Zen) - Simplicity and visual storytelling
- **Nancy Duarte** (Resonate) - Story structure and audience as hero
- **Guy Kawasaki** - 10/20/30 rule (10 slides, 20 minutes, 30pt fonts)
- **Seth Godin** - 6 words maximum per slide
- **TED** - Visual-first, minimal text, high impact
- **Edward Tufte** - Data integrity and information design
- **Steve Jobs** - Rule of Three and surprise moments

## Workflow: Guided Creation with Checkpoints

### Phase 1: Discovery & Context Analysis

**User provides**: Topic + Audience + Duration

**Skill analyzes**:
- Presentation type (board update, keynote, training, TED-style)
- Audience level (technical, executive, general public)
- Key message and desired transformation
- Story framework recommendation

**Checkpoint 1**: Present 3 structure options with rationale → user approves

### Phase 2: Content Development

**Skill develops**:
- Research topic if needed (web search, Obsidian knowledge)
- Extract maximum 10 core concepts (Kawasaki rule)
- Apply "what is / what could be" alternation (Duarte)
- Draft slide outline with one concept per slide
- Identify slides needing visuals

**Checkpoint 2**: Present outline + visual plan → user approves

### Phase 3: Visual Strategy

**Skill proposes** for each visual slide:
- Image type (infographic, photo, diagram, data visualization)
- Visual concept/metaphor
- Aesthetic direction (colors from brand, style, mood)
- Estimated generation cost ($0.13-$0.24 per image)

**Checkpoint 3**: Present visual strategy → user approves images to generate

### Phase 4: Generation & Validation

**Skill executes**:
1. Generate approved images in parallel using image-gen skill
2. Build slides using selected adapter (PPTX/Google Slides/Canva)
3. Apply design rules automatically:
   - 30pt+ fonts minimum
   - <6 words per slide (Godin standard)
   - High contrast (4.5:1 minimum)
   - No bullet points
   - No paragraphs
4. Run validation scoring (0-100 scale)
5. Generate quality report

**Checkpoint 4**: Present presentation + quality report → user reviews

### Phase 5: Iteration & Refinement (if needed)

**Skill refines**:
- Apply requested changes
- Re-validate quality score
- Save final presentation to Obsidian Research folder
- Document learnings and metadata

## Validation Scoring (0-100)

### CRITICAL Violations (Auto-fail)
- Font size < 30pt
- >10 core concepts
- Bullet points detected
- Paragraphs (>2 consecutive sentences)
- Poor contrast ratio (<4.5:1)
- Default template usage

### WARNING Flags
- >6 words per slide
- >15 slides for 20-min presentation
- No images in presentation
- Text-heavy slides (>3 lines)
- Inconsistent fonts (>2 families)
- Low-resolution images

### Quality Score Breakdown
- **Simplicity** (10pts): Word count, visual clutter
- **Visual Dominance** (10pts): Image quality, text-to-visual ratio
- **Story Structure** (10pts): Narrative arc, emotional beats
- **One Idea/Slide** (10pts): Concept clarity
- **Typography** (8pts): Size, consistency
- **Layout** (7pts): Hierarchy, whitespace, alignment
- **Color/Contrast** (7pts): Readability, brand consistency
- **Media Quality** (8pts): Image resolution, relevance
- **Cognitive Load** (20pts): Mayer's 12 multimedia principles
- **Data Integrity** (10pts): Tufte principles (if data present)

**Target Score**: 85+ for high-quality presentations

## Slide Pattern Library

### Six Core Patterns

1. **Title Slide** - Minimal text, strong visual, sets tone
2. **One Big Idea** - Single word/number/image, max 6 words
3. **Visual + Caption** - Large high-quality image, short caption
4. **Data Visualization** - Charts following Tufte principles (maximize data-ink ratio)
5. **Timeline/Process** - Linear flow, minimal text per step
6. **Transition/Section Break** - Single word or phrase, signals shift

### Adaptive Pattern Selection

**Board/Executive Update**: Patterns 3, 4, 6 (data-focused, professional)
**Keynote/TED-style**: Patterns 1, 2, 3 (story-focused, visual-first)
**Training/Education**: Patterns 4, 5 (process-focused, clarity)
**Pitch/Demo**: Patterns 2, 3, 4 (impact-focused, evidence-based)

## Story Frameworks

### Nancy Duarte's Sparkline
- Alternate "what is" (current reality) with "what could be" (aspiration)
- Build tension through contrast between present and future
- End with transformation and new reality

### The Rule of Three (Steve Jobs)
- Break presentation into 3 main sections
- 3 key features/points per section
- Memorable and dramatic

### TED Structure
- Hook (first 30 seconds) - Grab attention
- Personal connection - Why this matters to you
- Core idea with evidence - The meat
- Call to action - What next
- Strong close - Never end with Q&A

## Output Format Support

### Current: PPTX (Phase 1)
- Uses existing pptx skill
- Direct PptxGenJS API for simple slides
- html2pptx.js for complex layouts
- Saves to Obsidian Research folder
- Generates quality validation report

### Coming: Google Slides (Phase 2)
- Google Slides API v1
- Integrates with google-workspace skill
- Supports all 3 accounts (psd, kh, hrg)
- Returns shareable link

### Future: Canva (Phase 3)
- Canva REST API
- Auto-apply brand kits
- Returns shareable link

## Image Generation Integration

### Intelligent Visual Recommendations

For each slide, analyzes:
- **Numeric data** → Infographic recommendation
- **Abstract concepts** → Conceptual image/metaphor (especially for keynotes)
- **Concrete examples** → Photo/realistic imagery
- **Processes/flows** → Diagram recommendation
- **Trends/comparisons** → Chart/data visualization

### Prompt Generation (6-Step Process)

From image-gen skill:
1. Extract narrative from slide content
2. Choose visual concept/metaphor
3. Apply aesthetic (brand colors, presentation style)
4. Construct detailed prompt
5. Execute generation
6. Validate output

### Cost Management
- Show total estimated cost before generating
- Allow opt-out of specific images
- Cache generated images for reuse
- Track spending in presentation metadata

### Naming Convention
```
{presentation-id}-slide-{number}-{type}.png

Examples:
cybersecurity-2025-slide-02-threat-timeline.png
port-ai-2025-slide-04-breakthroughs.png
```

## Brand Integration

Automatically detects and applies branding:
- **PSD presentations**: Use psd-brand-guidelines skill (Sea Glass, Pacific, etc.)
- **Personal presentations**: Use personal brand (if defined)
- **Client presentations**: Custom brand kit (future)

Applies to:
- Color palettes
- Typography choices
- Logo placement
- Overall aesthetic

## Learning & Metadata Tracking

After each presentation, save metadata to Obsidian:

```json
{
  "presentation_id": "cybersecurity-2025-board",
  "type": "board-update",
  "audience": "school-board",
  "duration": "15min",
  "slides": 10,
  "quality_score": 91,
  "images_generated": 7,
  "generation_cost": 0.94,
  "validation_warnings": 1,
  "user_edits": ["slide-8-wording"],
  "effectiveness": "approved-minor-changes",
  "created": "2025-11-24",
  "tools_used": ["research", "image-gen", "pptx"],
  "learnings": "Board prefers data charts over concept graphics"
}
```

## Integration with Existing Skills

**Dependencies**:
- `image-gen` - Visual content generation
- `pptx` - PowerPoint creation (Phase 1)
- `google-workspace` - Google Slides (Phase 2)
- `psd-brand-guidelines` - Brand colors/logos
- `research` - Content development
- `obsidian-manager` - Knowledge storage

**Coordination Pattern**:
```
presentation-master (orchestrator)
├── calls research for content depth
├── calls image-gen for visuals
├── calls pptx/slides/canva adapter for output
├── calls psd-brand-guidelines for colors/logos
└── calls obsidian-manager to save final + metadata
```

## Usage Examples

### Example 1: Board Update
```
User: "Create presentation on cybersecurity trends for school board, 15 minutes"

Geoffrey: I'll create a board update presentation on cybersecurity trends.

**Context Analysis**
- Type: Board/Executive Update
- Audience: School board (non-technical)
- Duration: 15 minutes
- Purpose: Inform + Recommend action

**Recommended Structures**
1. Problem → Solution → Action (8 slides)
2. Current State → Trends → Implications → Recommendations (10 slides) ⭐
3. Story: Recent incident → Broader context → Our response (7 slides)

Which structure resonates?

[User selects #2]

**Draft Outline** (10 slides, one concept each)
1. Title: Cybersecurity Trends 2025
2. Current Threat Landscape [visual: threat timeline]
3. Trend 1: AI-Powered Attacks [visual: concept illustration]
4. Trend 2: Ransomware Evolution [visual: data chart]
5. Trend 3: Supply Chain Risks [visual: network diagram]
6. Impact on Education Sector [visual: stat infographic]
7. Our Current Posture [visual: assessment chart]
8. Gaps & Vulnerabilities [text-focused, high contrast]
9. Recommended Actions [visual: roadmap]
10. Investment & Timeline [visual: budget chart]

**Visual Strategy**
- 7 images recommended
- Estimated cost: $0.94 (2K resolution)
- Mix: 3 infographics, 2 charts, 1 diagram, 1 illustration

Approve and proceed?

[User approves]

Generating 7 visuals in parallel...
✓ All generated (0.94 total)

Building presentation with PSD brand colors...
Validating against best practices...

**Quality Score: 91/100**
✓ All fonts 36pt+
✓ Avg 4.2 words/slide
✓ High contrast (8.5:1)
✓ Story arc present
✓ Cognitive load optimized
⚠ Slide 8 has 7 words (rec: 6 max)

Created: /Obsidian/Research/cybersecurity-trends-2025-board.pptx
```

### Example 2: Keynote
```
User: "Create keynote on future of education for conference, 30 minutes"

Geoffrey: I'll create a TED-style keynote on the future of education.

**Context Analysis**
- Type: Keynote/TED-style
- Audience: Education professionals + parents
- Duration: 30 minutes
- Purpose: Inspire + Shift perspective

**Recommended Structures** (Nancy Duarte's Sparkline)
1. What education is today → What it could be → Transformation (12 slides) ⭐
2. Steve Jobs Rule of Three: Past → Present → Future (9 slides)
3. Hero's Journey: Problem → Trials → New reality (10 slides)

Which framework?

[Continues with guided creation...]
```

## Quality Checklist

Before presenting to user, verify:

**Content**
- [ ] One main idea per slide
- [ ] Clear story arc (beginning/middle/end)
- [ ] Audience positioned as hero
- [ ] Emotional connection established
- [ ] Maximum 10 core concepts

**Design**
- [ ] All fonts 30pt+ minimum
- [ ] High contrast (4.5:1+)
- [ ] No bullet points
- [ ] No paragraphs
- [ ] Maximum 6 words per slide (or justified exception)
- [ ] All images high-resolution
- [ ] Consistent fonts/colors/alignment
- [ ] No transitions or animations

**Data Visualization** (if applicable)
- [ ] Lie factor 0.95-1.05
- [ ] Maximum data-ink ratio
- [ ] Clear, detailed labeling
- [ ] No chart junk (3D effects, unnecessary borders)

**Structure**
- [ ] Follows chosen framework (Duarte/Jobs/TED)
- [ ] Creates tension and resolution
- [ ] Includes surprise/memorable moments
- [ ] Ends strong (not with Q&A)
- [ ] Appropriate length for time

## Advanced Features

### Context Detection

Automatically infers presentation type from cues:
- "board meeting" → Board update (data-focused)
- "keynote" or "conference" → Keynote (story-focused)
- "training" or "workshop" → Educational (process-focused)
- "pitch" or "investor" → Pitch (evidence-focused)

### Smart Defaults

Based on detected type:
- **Board update**: 10 slides, 8 with visuals, professional tone
- **Keynote**: 12-15 slides, minimal text, emotional arc
- **Training**: 15-20 slides, process diagrams, step-by-step
- **Pitch**: 10 slides (Kawasaki rule), data-driven, ROI focus

### Accessibility

Ensures presentations are accessible:
- High contrast text (4.5:1 minimum, aim for 7:1)
- Large fonts (30pt minimum, 36pt+ preferred)
- Clear hierarchy and flow
- Alt text for images (future enhancement)

## Limitations & Future Enhancements

**Current Limitations**:
- No animation support (by design - Godin principle)
- No video embedding yet (future enhancement)
- No speaker notes generation (Phase 2)
- No slide master editing (uses templates)

**Future Enhancements**:
- Auto-generate speaker notes from slide content
- Suggest rehearsal timing
- A/B test different structures
- Track presentation effectiveness scores
- Build personal pattern library from successful presentations

## References

Full principles documentation in:
- `principles/masters.md` - Expert best practices
- `principles/validation-rules.md` - Quality scoring system
- `principles/slide-patterns.md` - Template library
- `principles/story-frameworks.md` - Narrative structures

Scripts and adapters:
- `scripts/analyze-context.js` - Type detection
- `adapters/pptx-adapter.js` - PowerPoint output
- `adapters/slides-adapter.js` - Google Slides (Phase 2)
- `adapters/canva-adapter.js` - Canva (Phase 3)

---

**Remember**: The best presentations are simple, visual, story-driven, and focused on the audience's transformation—not the speaker's information.

© krishagel · MIT · krishagel/geoffrey

Similar skills

All skills for Design →

From the same set2

Project sponsors

Companies that keep this analytics open